VMware Velero Community Meetings, 2 Jul 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Velero Community Meeting - July 2, 2019

Description

In this meeting we cover the road to 1.1, and Carlisia does a great lengthy demo of several Velero features, check it out!

As always, for updated community info head over to https://velero.io/community and follow the links.

A

Hi everyone and welcome to the first Valero community meeting of the second half of 2019 super excited to have you all here. We've got the entire team. We're gonna, give you a good update and we invite you all to ask questions, of course throughout the meeting. So if you have any questions regarding the updates or the demos that we're gonna be showing today, please just raise your voice or send us a message in the chat if yeah, so this is being recorded and will upload it up on YouTube later on.

A

So let's get things started and I'll share my screen, so you can see the agenda as well.

A

So, first off, when we look at things that we've been working on, we got Nolan who's gonna, give us a little update on the CSI snapshot. Prototype plugins, yeah.

B

So I don't have a whole lot more to elaborate on with the prototype plugin I think the approach is mostly sound, mostly we're looking for people to give it a test against CSI implementations that have snapshot, support or are trying to add snapshot, support also in the meantime, I'm working on APR for getting rustic support for CSI volumes, because that was a missing feature and rustic support. So really, just looking for more people get looking to get a broader test test based on the plugins and see if we can suss out any more issues.

A

Cool any questions on the CSI plugin work Nolan's been doing if you missed the the demo. That Nolan did was that two weeks ago, it's up on YouTube as well. So that's in the previous community meeting. So if you want to see that head on over to YouTube- and you can see it there all right next up- we got kaliesha back from Cuba on Shanghai, hey.

C

Everybody's, yes, so Shanghai was great. The city is amazing: there were 3,000 I, ten G's there and.

C

He was, it was really good anyway, I am so I am so I came back, taking back half of the agrestic GA issues so, in other words, the issues that we are going to get done to consider drastic general family. If you want to see what they are, it's on that link that Jonas is showing on the video right now and if you have questions, if you want to get pick up any of the tickets, let me know you want to test anything. Let me know.

C

It's most mostly a performance, improvement and usability improvements.

C

C

All right, let's move on I, have a demo I'm going to do this demo, because why not? I did this demo for my presentation in shanghai and it is to show how to migrate from one cluster to another cluster.

C

So as you all most of you know, but for people watching this video, you might or might not know that the letter is towards the backups in the nest, three buckets and usually that buckets is in the same provider as the cluster you backing up from, but it doesn't have to be can be anywhere where you can set up identity buckets. You know, I'm going to show you set up where the bucket is not in a cluster. So let me share my screen.

C

How is the size fonts I.

A

Think you can make it a little bigger.

C

How about this that's.

A

C

It good enough.

D

Well, it could be a little bigger for me. Okay,.

D

That's better: it's.

C

A good enough, yes,.

D

That's good! Thank you. Okay,.

C

Sure, all right so before I show you the actual demo. Let me walk you through the setup that I have here, so you can understand. What's going on as I do things I have? Let me move so I have two clusters up in brogni and I'm using kinds so I'm going to put for you, my darker. What's what I have running here with my docker container?

C

There's so I have one cluster, so the top two containers are kind containers running the kubernetes cluster, so one I named development control, so development cluster and the other one I named staging cluster, and this is so. You can imagine a scenario where you have a development team or your developer. You open up an app and you are.

C

You have to ship that to may be a stage, an environment where people will test and you can be doing that multiple times a day once a day or once a week, and so the people who need the access to die stage. An environment will have an exact version of your application.

C

If you just do we usually return restore to their other customer, so here I have so like I said two containers, one for each of my clusters and I have a separate container running menial and I'm going to have both clusters pointing to that Mineo setup. So one is clean, it's the push backups to the menu pass, three buckets and the other one is going to pull from there and I also have here. This window is, if you see the bottom of my screen on the top right, you will indicate both the name space.

C

C

In the cut in and on this particular window, I dedicate this window for the development Custer and this window for my station cluster and I'm going to go forth between them, but I'm going to start here with my development cluster. So the first thing I'm going to show you is so I using K for a shortcut for keep control. So you know and I'm in the Valero namespace. What I wanted to show.

C

You actually was a get name space but spoiler I'm already in if a little knowing space but I want to show you is that filler is a very last name space there oh hold on. Let me delete something.

C

I'm deleting the app that I am going to start to show you. It is part of my demo, so the very last names there is the Vietnamese name space. That means Valero is in salt, for example, if I go on, my staging cluster and I say get name space link spaces you see, filler is not there. I'm going to install it later and I will just to demonstrate how easy it is through the software.

C

So going back to my the government cluster so that engine X name space is not there anymore great and another one I think I'm going to show you is that I have a bunch of scripts and they are just so I, don't say here, mistyping things and make you wait. Why are miss type things, but they are life they're going to be running life against the clusters that I have so.

C

The first thing that I want to do is not backup, because mainly I, don't have anything of interest in this in this cluster, so I'm going to add something of interest, I'm going to add an L, and that app is.

C

Okay, so that app is just a nginx server.

C

So basically, I have a manifest there in that directory is called base demo and so I created that, and that gave me namespace called nginx example. That's the namespace, where I put that that's container, so you can see it. There is the second one from the bottom and if I do get I'm going to switch to my.

C

In the next example, if I say get parts on this namespace you're going to see here, the pods has been created and it's running and it's ready great. So now. What I want to do is just to demonstrate that that is this really running in life and to do that. I'm going to do I have to add I.

C

Have to do a port forward on that server.

C

Yeah, okay, so I'm going to add port 8000 to that server. So we can access.

C

C

Don't think that was the that was the script. I run it no sorry this is the script. I run it here. We go great, so I'm gonna pop up back up here on this pane, clear, I'm gonna going to curl. Can you see that so I'm going to curl that port that I just opened- and you can see here- welcome to nginx? You can see down here that it's a nothing connection, so each time I think I get one entry Hey for that, it's great it's working its life. Alright!

C

So now, I have something of interest to that. I want to make sure I back up I'm going to do velaro, create backup and I'm going to give it a name. I'm going to say pack me there it could be. My company name could be my product name, my app name and, as you can see up here, we are getting two messages here. One is saying the backup request was sent successfully.

C

So this is just a big request to create the backup is not really the same. The backup is was successful so for us to check if the backup was created successfully. The other gives you these two instructions here and I'm going to first run the describe this is what I usually start with, because it's short, and so it's going to tell us if the backup was really created because it could fail validation.

C

It was fail to upload to the s3 buckets and in which, in those cases it could give you a failed face, or it could give you a take. Its partial failed, something that's that would translate to partially fails and it will give you the error messages on the bottom. Well, we don't, we didn't get any error, so that's great and so face completed. That's how I know we say so.

C

It's completed here and I'm going to run the logs I know: they're, not gonna, be our along, but just wanna show you what it will look like. So this is a log just for that back up, so we can't look at the log for the deployment switch which will combine every all the transactions, but you can also just look at particularly one like each individual backup blog, which I think is really neat alright.

C

So now, I have app running on the development cluster I have a backup for it, and I can do a little get backups and see all the backups that I have, and you can see here the name the stars is completed and the expiration timeline and the storage location. You can configure different storage locations. I'm not going to go into that. Of course, if you have questions about how to do that, you can have multiple locations.

C

If you need to set up your workflow like that, so you can ask us questions if you to learn more and so I'm going to go to my staging cluster and remember I, don't even have a little installed here. So the first thing I'm going to do is solve their own.

A

C

All right, so let me go over what well it just happens here.

C

Sevilla has this very neat install command where you pass these flags to set up everything that you need to set a filter on the cluster, so this is going to be the Valero server running on the on the cluster. The secret file is where you will put the credentials for your service provider, and so here's the credentials for my min us three buckets and the backup location and it's going to be the default that gets created but I'm specifying the region is me know just to give an indication that this is a menial set up.

C

We have a lot of information on our docks, and here something of interest is that the URL for the minibar kids I'm using my computer's IP address. If I use localhost, he won't be able to find it, because this is running minerals in the container and I'm trying to access menu from my closet. That's in another container and I I'm telling me telling it the lair of the port it's running on because that's the before the container is running on, and this is also really neat.

C

I'm passing a local image that I have here on my computer and hmm excuse me.

C

Neat, if you're just trying to add a little weight, an image that let's say you know in a network that doesn't have access to the outside world roads. You can just pull it from your local image and everything that gets set up is based on clg's.

C

So we create a CRC for each resources, its resource, that we need- and this is IC ed and bestowed hooray, and we can do a quick. We can run this command.

C

To to make sure that okay,.

C

C

Okay, here we go.

C

C

All right so right away the in stock cumin gives you the the log that you should be checking if any problem happens with installation. You will see here great now, I.

C

Have the name, space, Valero and I: don't have the nginx namespace, because I have installed that I up on this cluster and I'm, not going to what I'm going to do is my grades that workload the workload for my first class or into this cluster, and this is how I'm going to get access to that I'm, going to show you how to do that and I'm going to prove to you that it worked so hang on to your seats. Let me see what I want to do.

C

I wanna I wanna run a risk for my restore script, so.

C

Here, you're going to see how you create a restore and basically very simple, the lower restore, creates, that's a name that I pad that I gave to.

C

It is actually I'm just generating random names on mice with my scripts, but you can either pass so you could pass whatever name or you can pass no name in the letter or create a name for the restore and the relevant heating here is the from back up and I'm passing the name of the backup I want to restore from and because I have setup the backup, storage location, two points to the to the s3 bucket that I have running like that was connected with my first cluster.

C

So when I passed, this name Valera knows: okay, that's the backup that I need to restore from, and so it's again it says the request was submitted successfully. It doesn't. That was the request that was sent successfully. We don't know if the request worked so I'm going to run here. The describe is my favorite first thing to track and if I scroll up it's a face, completed perfect I'm, not even going to run the log.

C

It we know it works and now, if I say your name space as I expected my and Ning engine X example names there, that's a great sign. So I am going to switch to the name space, so I don't have to type so much.

C

Okay, so that and I'm going to do, get pods and see that my engine export is running and it's very fantastic. Now the restorer seems to have worked, but deliberately so I'm going to do the.

C

Let me do this I'm going to do a pop forward on that server and I have even a different scripts, because I didn't close the port on that first server, and so that port is v8. Thousands being used so I created four forward here with a the port 81 8 8, 1 0 0 anyway, so I'm going to pop up here and do a curl on the ports and check that out your worth.

C

I just did a restore it's like magic, I know and one more thing that I want to show I'm going to fix it. This is feign to make my room clear. This now falero yeah back up as we expected, even though I'm in the staging cluster, like because my bucket is pointing to the that the same as three buckets. That's my eye.

C

My first part, though development, but it was pointed to I- can see the bus so the the right way. One thing I'm not showing is that, with the with a little point, one point zero: we when we could create this this, the second cluster that we mainly wanted to restore from a different cluster. We would create it as read-only, so we we would prevent people from accidentally doing a backup of that factor into the f3 bucket, because it would be a mess.

C

That's not how things would work, so you would purposefully create it with a real bridge, only flagging, but so after 1.0. So our next release is going to have a very neat feature that you don't have to set the whole bucket as read-only, only the store location. So you, if you have a pockets, if you have a cluster here, like in my case, the development cluster and I'm backing up the cluster, so I set up.

C

I can have multiple storage locations which I didn't demonstrate, but we can in again you can ask us questions or without docks, so I said I said a backup to go to a specific store location. So that's going to be written right so when I set up this cluster over here to restore from that backup, location I can, instead of specifying the whole cluster to be read-only. I can specify only that backup, start storage location that way I can create another storage, location and backup that second cluster, because now it's not the cluster, that's real.

C

Only this is just be specific. There cut locations, I, don't know if there is a question. I see a chance. Okay, awesome demo, oh yeah, great all right John is this is very distracting. Tell me if there is any question, please feel free to interrupt me. I'll be glad. Okay, yes, yeah.

A

No, no questions for you. We do have a question for our own CSI, but yeah no question is we run the demo, except for awesome demos, great.

C

Thank you. So there is one last thing I want to show, and that is so. Let me backtrack a little bit, so we have that backup. I did not set either this cluster or the backup location as the read-only, because I am on going against master I could have done just the backup location. It's read-only, so I, so I could into I could technically delete. I can't delete this backup from here, because I didn't do that. It's really bad practice.

C

We have to be to make sure we don't do that and so just not to encourage bad practice. I'm going to go back to the cluster there I should have rhythm right permissions, which is my development, cluster and I'm. Going with oh, so two more things I'm sorry I lost a little bit of track. I'm going to show you two more things. One is how to download your backup and then I'm going to show you how to delete backup.

C

So the little has a very neat feature that if you have a backup on a cluster, you can do it downloads. So you do valet or download. If you say what it is that you want to download as a backup- and you say you pass the name of the backup and.

C

It might do this wrong. Oh my gosh, okay, hey I'm blanking. Somebody help me yeah.

E

I think I think you may have to do melero, backup download. We may not have this one configured to work in either order.

C

Okay: okay, thank you great now, I.

C

Am going to go the Oh all right? Sorry I'm going to do it again because I wanted to do it inside this directory, because I want to explain the fire and I didn't want to have all the scripts showing. So you can see so in this download directory, where I am now.

C

But see that I have the first fireworks and the second file is the turbo with all the backup contents. So.

C

I'm going to actually let me run the scripts because it has a command to.

C

To on tarp I, don't remember by head by memory, okay great, so that was just showing the the files and now I ran the commands to on tar, and here you can see all of the resources that we had in the cluster and if I expand this you see inside each folder, you have all of these items and bunch of stuff.

C

So here is our in the next example resource and it's you know it's on your local now is on my local computer.

C

Okay, so now again.

C

That's my backup still there in the last thing I want I want to show is that I can delete that backup.

C

Very conveniently in thank goodness there is a prompt asking me if I am sure. So again the request was sent so one thing that I can do to make sure that that backup was deleted. It's just you know, get backups and voila, it's not there anymore. Usually, if you have persistent volumes and a lot of things, this will take a little bit of time, but since I didn't have any persistent volume and it's a everything is running locally, it was super quick and that's it. That's the end of it.

A

Awesome, thank you. So much Galicia, you know great demo. We do have one question here regarding the demo from Taher. Do these downloaded backups contain persistent volume metadata.

C

Someone correct me if I'm wrong, but yes.

E

So yeah, so if you, if you had persistent volumes that were backed up like that, the yamo backed up as part of your backup, they would certainly be in there. If you had persistent volumes snapshots, we actually store the metadata about those snapshots in a different file which is and download it as part of this command, but I'm thinking. We, we may want to actually modify this command to download the additional files that have those other metadata so that you get all of the information about the backup when you're on this command.

F

Oh cool, very, very cool thanks thanks Steve, the the use case I was thinking about is, if you lost your storage and your cluster, but you have these snapshots somewhere. Can we bring the the config brings back the applications and somehow, if I have the data somewhere? Can I restore yep.

E

Yeah, that's a that's a great point and I think that's! You know you could certainly kind of reconstruct these by by hand, but I think we should look at modifying that command so that it does download all those metadata files so that you could you could go ahead and upload them all and and be able to restore. Thank.

D

You Steve sure.

E

Oh thanks for the question.

A

Yeah, thank you so much and no Taher. You had another question that you want answered as well. Do you want to speak up yeah.

F

This is a Taher wara I work for a storage company, pure storage I've been looking at Valero for the past 3 to 6 months. We support a CSI plug-in and data protection for storage from pure pure storage. My company is a feature set. We would like to add, and so we just started looking at Valero integration. It seems to be the the de-facto tool in the Kuban. It is volt today to do backup and migration, and things like that.

F

We just started considering writing the block, store, plug-in and the object store, plug-in and I'm I'm very happy to see that we are look looking at using the CSI industry. Standard snapshot support through Valero. The question I had earlier was so that takes the the CSI plug-in integration into Valero would take care of the block store plug-in. As per my naive knowledge about Valero goes. You still need to write from object or plug in separately if needed,.

E

Yeah I can take this one, since no one had a drop for another call. So that's that's basically right. So the the way we're thinking about it is that you know in the in the short to medium term. We want to add support for integrating with CSI snapshots, but we're not immediately gonna be dropping our own block store plug-ins. We actually rename them to volume, snapshot or plugins, but we're not going to be dropping support for those anytime soon.

E

You know down the road in the longer term, as the kind of the community and the user base moves over to using CSI, we may look at dropping support for our volume snapshot or plugins eventually, and just just using CSI, but in the in the medium term, will support both CSI integration, as well as the the bolero volume snapshot of plugins and then yeah. The object store, plugins, aren't really affected by this. So we'll continue to support that as the interface for integrating with with any additional object, storage systems.

F

Okay, thank you. I have lots of other questions, I'll reach out to the Valero community team, on slack yeah.

E

That sounds great.

A

Here from Jessie, and then we got one more question for for Khaleesi as well, so Jessie says that's good to hear the OCI plugins are nearly ready for release. Yeah we're just gonna.

G

Drop it there. Just I was very relieved to hear that, like yeah, we are dropping them, but not quite yet. So it was like bringing me back to the flashbacks of chasing docker with Solaris and just every release. The API is changing so yeah. This will get done and I'll release it and then.

E

G

Also have an object, store, plugin, which I think will be the first or at least you know beyond the entry one. So that's been interesting because you have to do this little juggle because you, so you do you deploy with that type and then it falls over until you can install the plugin. So there's a there's, a PR around possibly adding plugins to the install commands, which I am happy to help with. As soon as the dust clears on this first round to work on the plugins themselves,.

E

No okay, awesome, yeah and I. Think I definitely think you know updating the install command to be able to just install a plug-in during the install will make that user experience a lot better. So look forward to working on that awesome.

A

Thank you so much for that Alex. You had a question. Do you want to speak up or do you wanted me to read your question.

A

All right, I'll read it then so Alex had a question for Alicia here he caught the end of the bolero download, so he's wondering if you can do an inverse, so you have the the file download of the tarball downloaded. Can you use that to kind of uploaded back to bolero as a backup? No.

C

We cannot, that sounds pretty no, but so tom is our product manager and I wouldn't ask him. If he thinks is. We should encourage the opening, an issue for that yeah.

D

I'll, take it I'll. Take a note on that. That's good.

E

You can you know, you can certainly do it manually, but there is no CLI command to do it at the moment.

A

Oh, thank you. Thank you for that question. Alex. Alright, let's look at the agenda here.

A

I'm gonna show the agenda, so people know what I'm talking about as well sure desktop all right. So we do have a few things coming up as well. The.

A

A

Was this something you want to talk about? Steve yeah.

E

I can just cover it quickly. This is, you know, just a quick reminder that the best way to kind of see the team's current status is to look at the Zen hub board where we're currently working on the 1.1 release, and so you can see the scope for that in the release backlog pipeline, which is I, think just off to the right of your screen here, Jonas and so yeah didn't didn't really need to cover.

E

Anything in specific, in particular, just wanted to remind folks that this is kind of the best way to to see where things are.

A

Cool and the Zen hub is linked from Valerio as well, so you can. You can always check it out there.

A

Another thing that's coming up on July 18 we velaro will be hosted or presented rather during the CNC F webinar series. So Tom and Steve will present velaro talk about the project and talk about yeah. What's coming up in the next release, what's been what's been going on in the CNC F webinar, so we'll link to that as well. Will we'll share that in slack and on Twitter and pretty much everywhere, so you can all join in that's going to be on Thursday July 18th at 9:00 a.m. Pacific, so that's noon, Eastern 6:00 p.m.

A

Central, European Time! So yeah, please join me. If you can, that would be. That would be amazing. Next up we got a Tom. You want to talk a little about the Valero user survey, yeah.

D

So we are putting together a user survey. As you guys know, we've have an issue we've been running for a while, which is the next bullet down issue 13 to 27. Tell us about how you use Valera. We love to hear how people are using it, but we're gonna expand on that to do a user survey. The user survey we're hoping to announce at the CN CF webinar in a couple of weeks, will have it open for a while and then we'll compile all the responses and and get those back to the community.

D

It's somewhat on velaro, but it's also on just the importance of backup and restore for cloud native applications as well. So we'd love to have everyone look for that link and when, when it gets announced here a couple weeks and fill it out, we'd love to hear your kind of detailed responses to those those questions, so that we can better gauge the interest in certain features in the community and also kind of gauge where people's minds are at terms of data protection as it relates to cloud native.

D

So please join us in and participating in that and we will I guess. I had pushed all the responses back out to the community, for you guys to take a look as well.

D

So I also mentioned the the issue 1327. Please fill that out if I'd love to hear how you're using bolero- and you know otherwise, you know- join us on slack at the hashtag Valero and on the Cobra, a slack ask us questions and we will respond and get you the answers you need, or you know if you see something you like in Valero or would like improved in Valero, just drop us a note there to. Let us know how we can make things better, but we'd love to love to continue getting feedback from the community.

A

And that's it for me. Thank you, Tom yeah, the the feedback here from all of you who are using bolero or working on Valero. It's super important for us to know what what you're thinking as well. So please help us with that and well, as Tom said, we'll share all the responses and everything anonymized afterwards we'll make sure that you have all the data as well.

A

Do we have any other questions from the community before we wrap up.

A

All right, I want to thank you all for for joining today. It's been it's been a pleasure to have you all here. Thank you so much for for the questions. Thank you for the demo of kaliesha. Thank you for the updates, Valero team and yeah. As I said, this will be posted up on youtube. So if you missed anything, we'll post it up on YouTube later today, so you should have it in the Bolero slack channel later today as well.

A

So you can see it and with that, thank you all and hope to see you in two weeks have a good one.